Relationship Between Visceral Metastases and Survival in Patients with Metastasis‐related Spinal Cord Compression

Objective To investigate whether visceral metastases have a significant impact on survival in patients with metastasis‐related spinal cord compression (MSCC), and to determine the difference in prognosis between patients with and without visceral metastases. Methods Three institutional databases were searched to identify all patients who had undergone spinal surgery for spinal metastases between March 2002 and June 2010. Data on patient characteristics including pre‐ and post‐operative medical conditions, were collected from medical records or by telephone follow‐up. Survival data were obtained either from medical records or by searching a governmental cancer registry. Results The mean age of study patients was 59.6 ± 10.5 years (range, 18–84 years), of whom 102 were male and 67 female. The median and mean postoperative survival times were 7.0 ± 0.5 (95% CI 6.0–8.0) months and 12.6 ± 1.2 (95% CI 10.1–15.0) months, respectively, in all patients, being 5.0 ± 0.5 (95% CI 4.0–6.0) months and 10.8 ± 2.4 (95% CI 6.1–15.5) months, respectively, for patients with visceral metastases and 7.0 ± 0.8 (95% CI 5.4–8.6) months and 13.0 ± 1.4 (95%CI 10.3–15.6) months, respectively, for patients without visceral metastases (P = 0.87). These survival times did not differ significantly between groups. Multivariate Cox proportional hazard regressions showed that visceral metastases had no statistically significant association with survival (P = 0.277), whereas rate of growth of primary tumor (P = 0.003), preoperative Karnofsky performance status (KPS) (P < 0.001), change in KPS (P < 0.001), and Frankel grade (P = 0.091) were independent prognostic factors in the whole cohort (P = 0.005). Changes in KPS (P = 0.001) and major complications (P = 0.003) were significantly associated with survival in patients with visceral metastases, whereas rate of growth of primary tumor (P = 0.016), change in KPS (P = 0.001), and preoperative KPS (P < 0.001) were significantly associated with survival in patients without visceral metastases. Conclusions Visceral metastases do not appear to predict the prognosis of patients with MSCC; thus, more aggressive surgery should be considered in patients with MSCC who have visceral metastases. Additionally, prognostic factors differ according to visceral metastases status in these patients.


Introduction
R ecent treatment regimens have prolonged median survival time in patients with cancer, which has consequently led to a high frequency of metastatic spinal cord compression (MSCC) during the remaining lifetime of these patients. Approximately, 70% of patients with cancer develop spinal metastases 1,2 , 20% of whom develop neurological deficits [3][4][5] . Almost 10% of patients with MSCC choose to undergo surgical decompression with or without stabilization 4,6-8 , which can restore neurological function and improve their quality of life. However, it is not yet clear how to identify the patients who would benefit most from surgical treatment. It is generally accepted that life expectancy drives treatment regimens for spine metastases 9 . For example, decompressive surgery is generally not considered indicated in patients with life expectancies of less than 3 months 10 .
Some surgeons and radiologists have therefore established various prognostic scoring systems for predicting survival to help decide in selection of the most appropriate treatment strategy 11 . Unsurprisingly, because visceral metastases are considered to indicate the terminal stage in patients with cancer and their treatment is palliative rather than curative, visceral metastases have been regarded as one of the most important, and therefore commonly used, prognostic factors. This factor has therefore been incorporated into all of these scoring systems [12][13][14][15][16][17][18][19] .
However, recently published studies have reported significantly disparate findings concerning the effect of visceral metastases on survival. Arrigo et al. 20 reported that visceral metastases do not significantly influence survival after surgery in patients with MSCC. Chong et al. 21 investigated preoperative prognostic factors in 108 patients and showed that visceral metastases are not an independent prognostic factor despite the median survival of patients with visceral metastases at the time of surgery being 4.0 months and that of patients without visceral metastases 11.0 months. Therefore, there is controversy over whether visceral metastases are a prognostic factor in patients with spinal metastases.
The current study was performed with the goals of further identifying the role of visceral metastases in predicting survival time in patients with spinal metastases and determining the difference in prognosis between patients with and without visceral metastases.

Methods
T his study was approved by the hospital Ethics Committee. Three institutional databases were searched to identify all patients with spinal metastases and Tokuhashi score 9-15 between March 2002 and June 2010 14 .
The inclusion criteria for performing surgical interventions comprised intractable pain despite medication, rapidly progressive neurological deterioration, and evidence of clinical or radiographic instability.
The exclusion criteria comprised spinal metastases without cord compression, treatment by radiotherapy or revision procedures, operative procedure vertebroplasty or kyphoplasty only, life expectancy less than 3 months, and patients whose medical condition was considered too poor to tolerate surgery. Life expectancy was estimated on the basis of the revised Tokuhashi scoring system. Additionally, surgery was selected by mutual agreement between the surgeon and patient.
Survival data were obtained from medical records, by telephone follow-up, or searching a governmental cancer registry. The patients were divided into two groups according to whether they had visceral metastases. Patient characteristics, including preoperative and postoperative medical conditions, were collected from medical records or by telephone followup. Selected possible prognostic factors were analyzed, and each variable was categorized into two or three groups as follows: age (<65 vs. ≥65 years), sex (female vs. male), rate of growth of primary tumor (rapid vs. moderate vs. slow), preoperative and postoperative Frankel scores (A-C vs. D-E), other bone metastases (no vs. yes), preoperative and postoperative Karnofsky performance status (KPS) (10-40 vs. 50-70 vs. 80-100), number of involved vertebrae (solitary vs. multiple), pathological fracture (no vs. yes), metastasis site (cervical vs. non-cervical), serum albumin concentration (<35 g/L vs. ≥35 g/L), sphincter dysfunction (no vs. yes), and interval between developing motor deficits and surgery (≤5 vs. >5 days).
On the basis of findings reported by Tomita 12 , primary cancer types were categorized according to growth rate as follows: slow growth (breast, prostate, thyroid, etc.), moderate growth (kidney, uterus, etc.) and rapid growth (lung, colon, liver, gastric cancer, and other cancers).
Postoperative survival was defined as the time between the date of surgery and death or the latest follow-up. Neurological function was graded according to Frankel grade preoperatively and 4 weeks postoperatively (patients with Frankel D and E are able to walk). Time to developing motor deficits was defined as interval between deterioration of motor function and surgery. Deterioration of motor function was defined as a change of at least one Frankel grade.

Statistical Analysis
Mean values are reported as mean AE standard deviation and median values with range. The characteristics of the two groups were compared using the χ 2 or Student's t-test, and a two-tailed P <0.05 was considered to denote statistical significance. Univariate analysis of survival was performed using the Kaplan-Meier method and log-rank test. Variables significant at P < 0.01 in the univariate analysis were tested through a backward stepwise selection process for their independent effect on overall survival (OS). Rate ratios and their 95% confidence intervals (CIs) were computed, as were odds ratios and their 95% CIs. P < 0.05 was considered to denote statistical significance.

Prognostic Factors in Patients with Visceral Metastases
Preoperative Frankel score (P = 0.035), change in Frankel grade (P = 0.013) and KPS (P < 0.001), local relapse (P = 0.020), and major complications (P < 0.001) were potential prognostic factors according to univariate log-rank test ( Table 2). The multivariate Cox regression model identified change in KPS (P = 0.001) and major complications (P = 0.003) as the only variables that were independent predictors of OS (Table 3).

Discussion
C urrently, most published studies that have focused on assessing prognostic factors in patients with MSCC have failed to distinguish between patients with and without visceral metastases. To the best of our knowledge, this is the first study to investigate the impact of visceral metastases on OS and identify different prognostic factors according to visceral metastases status.
A randomized controlled study 10 and a meta-analysis 22 have found that surgery is superior to radiotherapy alone in terms of functional outcome, pain control, and OS. However, not all patients with MSCC benefit from undergoing a surgical procedure. Especially in patients with short survival times, post-operative complications may offset the intended benefits of surgery, or death may occur before wound healing or functional recovery.
In general, patients with very short survival times are not suitable candidates for decompressive surgery 10 . Therefore, means of accurately predicting survival time in patients with MSCC is currently an important topic to research. Various prognostic scoring systems for predicting life expectancy of patients with MSCC have been developed. The scoring systems reported by Tomita 12 and Tokuhashi 13,14 are the most representative and commonly used systems; both use visceral metastases as an important prognostic factor for survival in these patients.

Effect of Visceral Metastases on Prognosis
Understandably, development of visceral metastases, an indicator of more aggressive tumors, is usually regarded as denoting an advanced stage of cancer. Patients with visceral metastases tend to have shorter survival because of cancer progression 3 . Lei et al. 23 reported that visceral metastases have a significant impact on survival in patients with MSCC from lung cancer. Crnalic et al. 24 observed that visceral metastases have a detrimental effect on survival of patients with prostate cancer, the median survival of patients with visceral metastases being only 4 months, as compared twitho 10 months for patients without visceral metastases. Drzymalski et al. 25 found that the presence of additional metastases at the time of diagnosis of spinal metastases is independently associated with a shorter overall survival.
Surprisingly, our results differed substantially from those previously reported using the scoring systems of Tokuhashi 13,14 and Tomita 12 . Our results were conflicting in that patients with visceral metastases did not have a significantly shorter survival time than those with spinal metastases alone. However, this finding was in accordance with other previous reports. Sellin 26 reported that visceral metastases do not affect prognosis according to multivariate analysis, their univariate analysis showed that it was significantly associated with worse overall survival. Jiang 27 identified no significant effect of the absence or presence of visceral metastases on postoperative recurrence or survival. In another study by Sciubba 28 , the median survival of patients without visceral metastases was 28.0 months, compared with 17.4 months for those with visceral metastases. However, the results of our multivariate analysis were similar to those of Arrigo 20 and Chong 21 in showing no statistically significant difference between patients with versus without visceral metastases.
In addition, visceral metastases status reportedly has a similar impact on prognosis in patients with different primary tumor types. Zadnik 29 examined the relationship of visceral metastases to survival in patients with MSCC from breast cancer and found that the median survival for those without visceral metastases was 25.9 months, compared with 28.1 months for those with visceral metastases; this difference was not significant on Mantel-Cox testing. Chen 30 reported that visceral metastases had no statistically significant association with survival in patients with non-small-cell lung cancer and spinal metastases who underwent spinal surgery. The findings of Park et al. were similar 31 . In addition, Ju 32 demonstrated that visceral metastases had no statistically significant association with survival in patients with MSCC from prostate cancer and Bakker 33 found that they were not significantly associated with survival in patients with renal cell carcinoma. Walcott 34 found that the concomitant   presence of visceral lesions or multi-focal bony disease did not have prognostic significance in patients with breast cancer. Thus, the presence of progressive systemic disease should not be a contradiction to aggressive surgery, which is in agreement with previous reports by Walcott et al. 34 .
The explanation for our results is unclear. One possible explanation is that the presence of spinal metastases in itself denotes a more aggressive and advanced stage of cancer than the presence of visceral metastases. Thus, survival is equivalent for patients with and without visceral metastases. Another possible explanation is that advanced treatment strategies, such as targeted therapy, hormonal therapy, chemotherapy and stereotactic body radiotherapy, effectively control systemic metastases and significantly prolong the survival time of patients with MSCC. It is also possible that there was a bias in selecting patients for surgery, because  patients with visceral metastasis usually have lower performance scale scores, which can be considered a contraindication for surgery. Additionally, differences in stage at diagnosis may have influenced our results. Another possibility is that visceral metastases may not affect the prognosis of certain types of primary cancer and those types may have accounted for a larger proportion of our study cohort, which may in turn have influenced our results.

Difference in Prognosis between Patients with and without Visceral Metastases
In the current study, we did not find a correlation between the presence of visceral metastases and decreased survival. However, we found to our surprise that patients with and without visceral metastases have different prognostic factors. The factors influencing survival times of patients with visceral metastases were change on KPS and postoperative complications, whereas rate of growth of primary tumor, preoperative KPS, and change in KPS were significantly associated with survival in patients without visceral metastases. These findings are in agreement with previous reports that have demonstrated that KPS, neurological compromise, and primary cancer type are associated with decreased survival. However, we did no further analysis to determine why and how other prognostic factors affect survival, because we only aimed to investigate the correlation between visceral metastases and survival. Of course, we believe that identifying these prognostic differences is important for selecting optimal treatment.

Primary Tumor
The prognostic impact of type of primary tumor on survival of patients with MSCC has been reported previously [12][13][14] .
Favorable histologic types such as breast and prostate cancer are associated with better survival prognosis than other types, whereas survival of patients with lung cancer is extraordinarily poor 20,35 . In our study, the rate of growth of the primary tumor was a significant prognostic factor in the whole group and the group without visceral metastases. However, we did not determine whether the type of primary tumor influences survival of patients with visceral metastases.

KPS or Change in KPS
In our study, change in KPS 4 weeks postoperatively was a significant prognostic factor in the whole cohort, as well as in both groups with and without visceral metastases. Preoperative KPS was a significant prognostic factor in the whole cohort and the group without visceral metastases, but not in the group with visceral metastases. One possible explanation is that poor preoperative KPS, or no or worsening KPS, denotes a more aggressive cancer or more advanced stage. Additionally, major complications may affect the prognosis of patients with visceral metastases. However, few studies have reported the prognostic impact of complications in patients with MSCC. In our study, the findings concerning influence of complications on prognosis may be questionable because of the large difference in the number of patients with or without complications (15 vs. 154). Further study is required to better address this question.

Limitations of the Study
First, this was a retrospective review with a small number of patients with visceral metastases, the small number possibly being attributable to financial considerations and the negative attitude of Chinese people toward seeking surgical treatment, especially for patients with visceral metastases. Second, a wide variety of primary tumors were included and different tumor types may have different biological behavior and different prognoses. It would likely be useful to analyze prognosis for individual tumor types rather than grouping all tumor types together. However, Abouret et al. 36 reported similar results in that they found that visceral metastases were not significantly predictive of long-term survival for various primary tumors. Third, in the present study we did not investigate the effect of chemotherapy because previous chemotherapy regimens varied between patients; those variations may have influenced survival. Last, we found it difficult to decrease heterogeneity between the two groups. Nonetheless, we believe that our findings are valid. Additionally, selection bias is inevitable in retrospective cohort studies 37 . In summary, visceral metastases had no statistically significant association with survival in patients with MSCC; thus, more aggressive surgery should be considered for patients with visceral metastases.